National Security - Page 4

Anthropic has published a new Usage Policy, taking effect on 12 November, with a section on deceptive campaigns, tighter wording on weapons and surveillance, and a ban on extreme abuse of its models. The UK's ICO says ten AI developers have made or committed to data protection changes, and has opened a call for evidence on AI agents. What changes, and for whom.

OpenAI says it banned two influence operations, one from Russia running a fake Latin American think tank and one from Iran pitching articles under seven fake bylines. It rates the Russian one the first "Category 5" operation it has disrupted. What happened, what AI did and didn't do, and what remains unproven.

Google Cloud has announced the Gemini agent, a single agent for knowledge work, media and code that can spin up sub-agents and "coworker" agents with their own identities. The Financial Services and Legal versions are in preview; Google gave no general-availability date. What it does, how Google says it is fenced in, and what is still unproven.

Three agent-security stories from 8 October. Zenity says one prompt to an exposed AWS AgentCore agent could lead to control of every agent in the same account and region, and AWS disputes it. Claude Code adds an option that blocks an action when a safety hook fails. Cisco Talos finds malware that writes to AI analysts. What each shows, and what it does not.

Manus raises more than $500m after Beijing blocked Meta's takeover. Tab emerges at a $300m valuation. Nvidia reportedly discusses another $1bn for Figure. ElevenLabs says it has passed a $600m run rate, and Synthesia launches Syren in beta. A Reuters/Ipsos poll finds most Americans think government is not taking AI risks seriously enough.

Anthropic has expanded its Cyber Verification Program into three tiers, Defense, Red Team and Specialized, that relax Claude's cyber safety blocks for vetted security teams on Opus 5.5, Sonnet 5.5 and Mythos 5.1. Specialized applicants are reviewed with the US government. What each tier allows, and what Anthropic's own test shows.

Three fired OpenAI safety researchers have reportedly urged the board to preserve the ability to monitor AI reasoning. Australia, meanwhile, has proposed making frontier AI firms prove their safety systems work, on a banking-and-aviation model. Both reported; one is a letter, the other a proposal.

Anyone can now check an image, video or audio file for SynthID watermarks from Google and partners including OpenAI, NVIDIA and Kakao, with Apple to follow. A useful check for some AI content, not a general AI detector. What a "no" result does and does not mean.

Anthropic's Claude Haiku 5.5 costs around 75% less to run than Haiku 4.5, by Anthropic's estimate, and jumps from 15.7% to 72.4% on Anthropic's OSWorld 2.1 computer-use test. Sonnet 5.5 cache reads are also halved. What is confirmed, what is company benchmark, and who it is for.

OpenAI is rolling GPT-6 into ChatGPT's Chat tab for every plan: GPT-6 Sol for paid tiers from 7 October, GPT-6 Luna for Free and Go from 8 October, plus "Intelligent UI" answers with charts, forms and mini tools. What changed, which Sol this is, and what OpenAI's own safety card flags.